Just days after most of Europe united under a single currency, microarray scientists from all parts of the continent, along with Americans and Britons, came together in Zurich, Switzerland, to exchange a different kind of precious currency: information.
The three-day conference organized by Cambridge Healthtech Institute, ”Lab-on-a-chip and Microarrays for Post-Genome Applications,” included a smattering of talks on protein chips, microfluidic microarrays, diagnostic and clinical research applications of microarrays, as well as novel technologies.
Overall, the talks evidenced a general trend away from basic technology explanations so often prevalent at conferences toward presentations about how different groups are striving to apply the technology in their basic science research and trying to move it into the diagnostic arena.
This new direction likely reflects the natural evolution of post-genomic era research. But it also could be a response to the plethora of microarray gene expression data that has been generated in the past couple of years, and the paucity of actual biological theories to assemble gene expression data into a meaningful whole.
“There are a lot of articles showing a lot of data, but [as to] what that data means and how the biological pathways work, there’s not much,” noted a conference chair, Tim Kievits, CEO of PamGene.
Later, when discussing applications of arrays in testing contamination of blood components, Jural Petrik, head of microbiology R&D at the Scottish National Blood transfusion service, agreed. “We are a long way from understanding the biology of [gene expression],” he said. “Until we understand the actual effect that gene expression has on the organism, the technology won’t be useful.”
Kievits added that he had “no doubt” that microarrays would eventually be used for testing samples in clinical labs. But he said, “the question is, which technology will cross the chasm” between aficionados of technology, and the slightly technophobic majority. “Only a few will make it.”
Austrian Pharma Group Switches from Incyte to Affymetrix
Interestingly, Kievits’ opening was followed by a presentation in which Christian Stratowa, a bioinformatics scientist at Boeringher Ingelheim Austria, discussed a pilot study using microarrays to look at gene expression profiles in B-cell chronic lymphocytic leukemia and correlate these profiles with clinical data. While the group initially sent their samples to Incyte to do hybridizations on cDNA arrays, Stratowa noted that his group has now switched to Affymetrix arrays, one of the few products that has seemingly crossed that chasm between the handful of technophiles and the rest of the scientific population.
The reason that the Boeringher Ingelheim group switched, said Stratowa, was that the biologists wanted to do the hybridizations themselves, and they found there were inherent difficulties in spotting arrays themselves, such as verification of PCR clones. They also found that 20 percent of the MAGE clones they had ordered from Incyte were wrong, while Affymetrix data was reliable.
Another problem that Stratowa’s group encountered had to do with normalization among different groups. The data from two different groups did not fit into a linear distribution curve — as many have discovered in microarray analysis — so the group had to develop a non-linear regression algorithm.
One reason for the data’s non-linearity may be that some of the leukemia samples were up to 10 years old, a fact that other researchers said could have interfered with the reliability of the data, as RNA tends to degrade even when frozen.
The group nevertheless found a number of genes that were correlated with survival rates, and 335 genes that were differentially expressed in different stages of the disease, and is currently conducting follow-up studies.
Struggling with Protein Chips
While studies using microarrays to correlate gene expression with clinical outcomes have become commonplace, pharmaceutical researchers are now trying to get one step further by developing protein chips that can serve as biomarkers for disease, drug efficacy, and toxicity.
“You can’t focus just on genetics and genomics,” said Sara Mangialaio, the laboratory head of immunoassay development at Novartis Pharma in Basel, Switzerland.
“You have to focus on proteins, because they undergo post-translational modification and protein levels can’t always be identified through measurement of mRNA.”
Mangialaio presented research on her effort to develop a prototype protein chip for cytokines related to rheumatoid arthritis. She chose to focus on this disease because of its prevalence, which she estimated at one percent of the Caucasian population. “This means two things,” said Mangialaio, “a big available sample, and a big market.”
Mangialaio and her colleagues at Novartis built the chip by immersing the substrate in a monomer solution, and building a self-assembled monolayer — a network of bristles that would prevent proteins from binding non-specifically to the surface of the substrate. They also tried both contact- and non-contact printing methods, and found that non-contact piezoelectric methods worked better. Next they used the increasingly popular “sandwich immunoassay” to capture the sample. In this process, biotinylated antibodies attach to the monolayer, and then the sample attaches to the antibodies. Streptavidin tagged with Cy5 dye is then added, binding to the sample as well, and creating a sandwich where the sample is the middle layer. To detect the binding, they scanned the chips with a laser scanner.
While this protein chip worked in principle, it still posed practical problems for the Novartis group. “We were still not able to quantify [one analyte] because the desired normal concentration is below the limit of detection,” Mangialaio said. In other words, the chip would have to be made more sensitive in order to differentiate between normal concentrations of arthritis-related cytokines and abnormally high ones. The coefficient of variation between chips was also 20 to 30 percent, and the group found that capture probes would become saturated at the high end of the analyte concentration.
Despite these difficulties, Mangialaio and her group are continuing with the project. They plan to use a Luminex bead platform to see if it eliminates some of the problems with sensitivity and variability, and also are looking at automating the protein deposition process on a chip platform. “We are convinced that comparison of healthy and diseased individuals using [protein] chips could lead to identification of disease markers,” she said.
Protein Chip Positioning
Two presentations that followed Mangialaio’s talk offered possible solutions for the problems of sensitivity and variability in protein chips.
Michael Saul, CEO of Piscataway, NJ-based BioArray Solutions, discussed his bead array technology in which up to 4,000 encoded beads are spread out onto a 300-micron-wide chip. Like other bead platforms, this one involves attaching the probe to a color-coded bead, and then detecting probe-target attachment with fluorescence. But the deposition of the beads on the chip, made from semiconductor materials, allows the signal to be detected with a standard microscope rather than a specialized instrument, said Saul. He also claimed that these beads, used as protein capture agents, would offer greater or equal sensitivity to that of an ELISA kit.
Steffen Nock, senior director of biochemistry at Zyomyx, also discussed the Hayward, Calif., company’s chip design and surface chemistry. To reduce between-chip variability Zyomyx has developed a PDC-dispensing robot that can make up to 1,000 chips per day using a non-contact process. While the company’s current chip chemistry involves the biotin-streptavidin sandwich assay, Nock said the company is looking at “label independent assays such as surface plasmon resonance, so we won’t have to put on the second antibody.” The company also plans to come out with a cytokine chip within the year, and is actively seeking partners to serve as beta testing sites.
Following these and other talks on protein chips, Thomas Joos, head of the biochemistry department at the University of Tubingen’s Natural and Medical Sciences Institute, discussed the inherent challenges and limitations of protein microarrays. The central issue, he stressed, is the intrinsic problem of miniaturization. Given that the amount of analyte is a constant, shrinking down the capture platform necessarily reduces the number of capture probes per analyte molecule, he said. Unlike DNA arrays, where oligos can be built on the surface like slim skyscrapers in a crowded city, protein capture probes cannot be squeezed together beyond a certain point where they start to become too dense to bind to the target analyte reliably. “If you understand this, you’ll understand the limitation of microarray technology,” Joos said.
This size limit is accompanied by other difficulties specific to protein chips, such as the inability — so far — to store protein chips for long periods. “For the diagnostic market, [the chip] has to have the capability of two years storage. We are not yet there.”
Even so, he said, the diagnostic market is not rushing to make a sturdier protein chip because it is not yet convinced it can make more money with protein chips than with microtiter plates.
But if protein microarray technology is developed further, Joos said, it offers “a lot of possibilities, more than just antibody-antigen interaction.” Other uses include detection of protein-protein interaction, autoimmunity, sandwich assays, and antibody profiling for affinity and specificity.
In a panel discussion that focused on protein chips, the overall consensus was that protein arrays, while needing improvements and refinements before they gain wide acceptance the way DNA arrays have, will find their time and place in the toolbox.
“DNA arrays took 10 years to take off, said Jeorg Hoheisel of the Deutsches Krebsforschungzentrum. “So give protein chips five years.”
Coming Next Week: Zurich conference report, Part II